Odd Solar Reading

[mcrossley]

At 02:03 this morning MX dumped a theoretical solar max of 892 into my database. Very odd, not seen this before. All the other values in that row of the MySQL table are normal.

A one off glitch?

Capture.PNG

[steve]

Looking at your graph, you also had a glitch yesterday at about the same time. Perhaps your system clock does something odd at that time, briefly? You have Cumulus running all the time, presumably - no restarts?

[mcrossley]

No restarts, I'll have a look at the rPi and see if if the logs show anything, it's just a standard headless Debian build. The monthly logs show nothing out of the order at 2:00 and 2:05.

F1 about to start so I'll take a look later...

[steve]

You could also have a look in the diags file to see if there's an odd timestamp in there. But it would depend on whether it had anything to log at that time.

[mcrossley]

Hi Steve, I had already looked in the diags file, nothing of interest in there.
None of the rPi logs show anything either, though the NTP service isn't set to log.
The date/time MX applied to the log entry is correct, and i checked it was logged at the correct time - i.e. between the 2:02 and 2:04 - and not put in position by a sort on the SQL query.
Yes, odd the same thing happened at the same time yesterday.
I'll switch debug logging on and see it it shows anything tonight...

[mcrossley]

Well, it hasn't recurred for the last two days, put it down to the cosmic faeries.

[steve]

Solar flares...

[mcrossley]

Another one last night at 00:01 - again nothing I can see in the logs - MX or Linux, the ntp daemon should abort and log into syslog if the time difference is greater than 1000 seconds (or that is my reading of the manual).

I have enabled ntp statistics logging now - but as this happens so infrequently...

[steve]

Yes, I think it's probably not anything to do with the clock going wrong. But what it is, is a mystery. One thing I have noticed, which seems to apply to yours and the newly reported case, is that it appears to be calculating the figure for the time 12 hours before (or after) the actual time.

[BeaumarisWX]

Hi Mark and Steve,
In case it is of any relevance, https://cumulus.hosiene.co.uk/viewtopic.p ... 76#p121376
2 (pi) users seeing same issue (Mark(UK)/Phil(AU)), my Win 10 has not displayed this (only mentioned this as Phil who also has experienced same as Mark is Australian based like me) .

Marks at UK 00:01 and Phil's at AU 04:31. (so no time relevance)
Regards,

[mcrossley]

Steve, my 'spike' from last night as a reading of 1084, currently I'm only hitting around 260 at midday, so I think that blows the 12 hour theory out?

Phil's profile says he is running Windows 7, if that is correct, then this not unique to the rPi either.

[BeaumarisWX]

Apologies Mark/Steve/Phil,
I had Phil mixed up with another user I was recently in correspondence with, my mistake. Thought Phil was (pi).
regards,

[Phil23]

Phil's profile says he is running Windows 7, if that is correct, then this not unique to the rPi either.

Actually the Weather laptop is on 8.1.

Does a scheduled restart at 4:00am each morning & Cumulus is correctly shut down at that time.

Only other change I've made is a registry entry to have it Poll the NTP server every 24 hours instead of the once a week default.

Phil.